Evaluation of various estimators for standardized mean difference in meta‐analysis

Meta‐analyses of a treatment's effect compared with a control frequently calculate the meta‐effect from standardized mean differences (SMDs). SMDs are usually estimated by Cohen's d or Hedges' g. Cohen's d divides the difference between sample means of a continuous response by the pooled standard deviation, but is subject to nonnegligible bias for small sample sizes. Hedges' g removes this bias with a correction factor. The current literature (including meta‐analysis books and software packages) is confusingly inconsistent about methods for synthesizing SMDs, potentially making reproducibility a problem. Using conventional methods, the variance estimate of SMD is associated with the point estimate of SMD, so Hedges' g is not guaranteed to be unbiased in meta‐analyses. This article comprehensively reviews and evaluates available methods for synthesizing SMDs. Their performance is compared using extensive simulation studies and analyses of actual datasets. We find that because of the intrinsic association between point estimates and standard errors, the usual version of Hedges' g can result in more biased meta‐estimation than Cohen's d. We recommend using average‐adjusted variance estimators to obtain an unbiased meta‐estimate, and the Hartung‐Knapp‐Sidik‐Jonkman method for accurate estimation of its confidence interval.     

The psychometric properties of the Migraine-Specific Quality of Life Questionnaire version 2.1 (MSQ) in chronic migraine patients

Objective

The Migraine-Specific Quality of Life Questionnaire version 2.1 (MSQ) has been shown to have good psychometric performance in measuring headache impact in migraine patients, but its properties specifically in chronic migraine (CM) patients are unknown. The objective of this study was to evaluate the psychometric properties of the MSQ in a group of CM patients undergoing prophylactic treatment.

Methods

Measurement properties of the MSQ were examined using two international, multicenter, randomized clinical trials evaluating onabotulinumtoxinA as headache prophylaxis in CM patients (N = 1,376). Confirmatory factor analysis (CFA) was used to test the latent structure of the MSQ in CM patients. The reliability, convergent and discriminant validity, and responsiveness of the MSQ were assessed.

Results

CFA confirmed the currently proposed three-factor MSQ latent structure across the two studies. Good reliability was observed for all three MSQ scales, across studies and time points. MSQ scale scores strongly correlated with the scores of the Headache Impact Test-6 (HIT-6). Analysis of known-groups validity indicated that MSQ scale scores discriminated between groups of patients differing in their 28-day headache frequency were as follows <10, 10–14, and ≥15 days, and the sample-derived quartiles of the total cumulative hours of headache were as follows <140, 140 to <280, 280 to <420, and ≥420 h (p < 0.0001), across both studies and time points. MSQ change scores were higher in magnitude in groups experiencing greater decline in headache frequency (p < 0.001).

Conclusion

The MSQ is a psychometrically valid tool that can be used to reliably measure the impact of migraine among CM patients.     

Estimating causal effects for multivalued treatments: a comparison of approaches

Interventions with multivalued treatments are common in medical and health research, such as when comparing the efficacy of competing drugs or interventions, or comparing between various doses of a particular drug. In recent years, there has been a growing interest in the development of multivalued treatment effect estimators using observational data. In this paper, we compare the performance of commonly used regression‐based methods that estimate multivalued treatment effects based on the unconfoundedness assumption. These estimation methods fall into three general categories: (i) estimators based on a model for the outcome variable using conventional regression adjustment; (ii) weighted estimators based on a model for the treatment assignment; and (iii) ‘doubly‐robust’ estimators that model both the treatment assignment and outcome variable within the same framework. We assess the performance of these models using Monte Carlo simulation and demonstrate their application with empirical data. Our results show that (i) when models estimating both the treatment and outcome are correctly specified, all adjustment methods provide similar unbiased estimates; (ii) when the outcome model is misspecified, regression adjustment performs poorly, while all the weighting methods provide unbiased estimates; (iii) when the treatment model is misspecified, methods based solely on modeling the treatment perform poorly, while regression adjustment and the doubly robust models provide unbiased estimates; and (iv) when both the treatment and outcome models are misspecified, all methods perform poorly. Given that researchers will rarely know which of the two models is misspecified, our results support the use of doubly robust estimation. Copyright © 2015 John Wiley & Sons, Ltd.     

Personal air sampling and risks of inhalation exposure during atrazine application in Honduras

Purpose

To assess occupational inhalation exposure to the herbicide atrazine during pesticide application in a developing country.

Methods

Personal air samples were collected during atrazine application using a personal sampling pump equipped with an OSHA Versatile Sampler (OVS-2) sorbent tube. Samples were collected from 24 pesticide applicators in Honduras. Application was observed during sampling, and a survey was completed in the home.

Results

Fourteen of the 24 participants used pump backpack sprayers to apply atrazine and 10 used tractor/boom systems. Despite applying about 15 times as much atrazine, the tractor/boom participants (11.5 μg/m³) had only slightly higher (not statistically significant) time-weighted averages (TWA) than participants using backpack sprayers (9.6 μg/m³). Within the backpack sprayer group, those that used a cone spray nozzle (11.54 μg/m³) had nearly double the TWA than applicators using a flat spray nozzle (5.98 μg/m³; P = 0.04). In the tractor/boom group, the participants that rode on the boom or the back of the tractor monitoring nozzles (15.0 μg/m³) had almost double the average TWA than tractor drivers (8.0 μg/m³; P = 0.097).

Conclusions

Since tractor/boom pesticide application decreases the number of man-hours required to apply pesticides, and does not increase inhalation exposure significantly, it decreases the overall population occupational exposure. Monitoring nozzles on booms from a distance rather than on the back of a tractor or boom may decrease or eliminate inhalation exposure. Use of flat spray nozzles for herbicide application among pump backpack sprayers may reduce their inhalation exposure.     

Improving weight status in childhood: results from the eat well be active community programs

Objectives

The eat well be active Community Programs (ewba) aimed to prevent obesity among children aged 0–18 years in two Australian communities from 2006 to 2010.

Methods

ewba was a multi-strategy intervention in children’s settings. The evaluation was quasi-experimental, including a before and after survey with intervention (INT) and non-randomised comparison (COMP) communities. Outcome measures included BMI-z score (zBMI) and overweight/obesity prevalence in children aged 4–5 years; and zBMI, waist circumference (WC) z-score and overweight/obesity prevalence in children aged10–12 years.

Results

After 3 years, among the 4–5 years old, mean zBMI was significantly lower in both INT (?0.20, p < 0.05) and COMP (?0.15, p < 0.05), however, changes were not significantly different between INT and COMP. There was a larger reduction in overweight/obesity prevalence in INT (?6.3 %) compared to COMP (?3.7 %) (p < 0.05, χ ² test). In the 10–12 years old, mean zBMI did not change significantly in INT or COMP. There was a significant reduction in WC z-score in INT (?0.17, p < 0.05) but not in COMP (?0.10, p = NS), although not significantly different between INT and COMP (p = 0.092).

Conclusions

These findings suggest that the ewba community intervention had a moderate impact, showing modest improvements in weight status at 3-year follow-up.     

Modelling spatial disease rates using maximum likelihood

This paper concerns maximum likelihood estimation for a generalized linear mixed model (GLMM) useful for modelling spatial disease rates. The model allows for log-linear covariate adjustment and local smoothing of rates through estimation of spatially correlated random effects. The covariance structure of the random effects is based on a recently proposed model which parameterizes spatial dependence through the inverse covariance matrix. A Markov chain Monte Carlo algorithm for performing maximum likelihood estimation for this model is described. Results of a computer simulation study that compared maximum likelihood (ML) and penalized quasi-likelihood (PQL) estimators are presented. Compared with PQL, ML produced less biased estimates of the intercept but the ML estimates were slightly more variable. Estimates of the other regression coefficients were unbiased and nearly identical for the two methods. ML estimators of the random effects standard deviation and spatial correlation were more biased than the corresponding PQL estimators. The conclusion is that ML estimators for GLMMs cannot be expected to perform better than PQL for small samples.     

The association between post-traumatic stress disorder symptoms and the quality of life among Wenchuan earthquake survivors: the role of social support as a moderator

Objective

To examine the role of the three types of social support as possible moderating factors between post-traumatic stress disorder (PTSD) and its relationship to two domains of the quality of life (QOL).

Methods

A cross-sectional survey was done in a local area near the epicenter of the severe earthquake in Wenchuan. The Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), the standard Chinese 12-item Short Form (SF-12v2), and the Social Support Rating Scale (SSRS) were used to interview a total of 2,080 earthquake survivors in the one-year follow-up period. Multiple regressions were performed to evaluate the moderating role of social support on the relationship between PTSD and QOL.

Results

Among survivors one-year after the Wenchuan earthquake, being a woman (p < 0.01), having a lower level of education (p < 0.01), having a lower level of income (p < 0.01), having a worse housing status (p < 0.05) and having a higher level of exposure (p < 0.05) were risk factors for a poorer QOL (?R ² = 0.063). PTSD symptoms negatively influenced the QOL (?R ² = 0.145), while social support positively influenced the QOL (?R ² = 0.016). However, the interaction between social support and PTSD weakened the negative effect of PTSD on the QOL (?R ² = 0.012). Subjective support and support availability moderated the association between PTSD and the QOL (?R ² = 0.010).

Conclusion

Subjective support and support availability are more useful strategies to improve the QOL of the earthquake survivors with PTSD symptoms.     

Association between Ambient Noise Exposure and School Performance of Children Living in An Urban Area: A Cross-Sectional Population-Based Study

Most of the studies investigating the effects of the external noise on children’s school performance have concerned pupils in schools exposed to high levels due to aircraft or freeway traffic noise. However, little is known about the consequences of the chronic ambient noise exposure at a level commonly encountered in residential urban areas. This study aimed to assess the relationship between the school performance of 8- to 9-year-old-children living in an urban environment and their chronic ambient noise exposure at home and at school. The children’s school performances on the national standardized assessment test in French and mathematics were compared with the environmental noise levels. Children’s exposure to ambient noise was calculated in front of their bedrooms (L_den) and schools (L_Aeq,day) using noise prediction modeling. Questionnaires were distributed to the families to collect potential confounding factors. Among the 746 respondent children, 586 were included in multilevel analyses. On average, the L_Aeq,day at school was 51.5 dB (SD= 4.5 dB; range = 38–58 dB) and the outdoor L_den at home was 56.4 dB (SD= 4.4 dB; range = 44–69 dB). L_Aeq,day at school was associated with impaired mathematics score (p = 0.02) or impaired French score (p = 0.01). For a + 10 dB gap, the French and mathematics scores were on average lower by about 5.5 points. L_den at home was significantly associated with impaired French performance when considered alone (p < 10⁻³) and was borderline significant when the combined home-school exposure was considered (p = 0.06). The magnitude of the observed effect on school performance may appear modest, but should be considered in light of the number of people who are potentially chronically exposed to similar environmental noise levels.     

Regression calibration for models with two predictor variables measured with error and their interaction,using instrumental variables and longitudinal data

Regression calibration provides a way to obtain unbiased estimators of fixed effects in regression models when one or more predictors are measured with error. Recent development of measurement error methods has focused on models that include interaction terms between measured‐with‐error predictors, and separately, methods for estimation in models that account for correlated data. In this work, we derive explicit and novel forms of regression calibration estimators and associated asymptotic variances for longitudinal models that include interaction terms, when data from instrumental and unbiased surrogate variables are available but not the actual predictors of interest. The longitudinal data are fit using linear mixed models that contain random intercepts and account for serial correlation and unequally spaced observations. The motivating application involves a longitudinal study of exposure to two pollutants (predictors) – outdoor fine particulate matter and cigarette smoke – and their association in interactive form with levels of a biomarker of inflammation, leukotriene E4 (LTE ₄, outcome) in asthmatic children. Because the exposure concentrations could not be directly observed, we used measurements from a fixed outdoor monitor and urinary cotinine concentrations as instrumental variables, and we used concentrations of fine ambient particulate matter and cigarette smoke measured with error by personal monitors as unbiased surrogate variables. We applied the derived regression calibration methods to estimate coefficients of the unobserved predictors and their interaction, allowing for direct comparison of toxicity of the different pollutants. We used simulations to verify accuracy of inferential methods based on asymptotic theory. Copyright © 2013 John Wiley & Sons, Ltd.     

Are community-living and institutionalized dementia patients cared for differently? Evidence on service utilization and costs of care from German insurance claims data

Background

The validity of survey-based health care utilization estimates in the older population has been poorly researched. Owing to data protection legislation and a great number of different health care insurance providers, the assessment of recall and non-response bias is challenging to impossible in many countries. The objective of our study was to compare estimates from a population-based study in older German adults with external secondary data.

Methods

We used data from the German KORA-Age study, which included 4,127 people aged 65–94 years. Self-report questions covered the utilization of long-term care services, inpatient services, outpatient services, and pharmaceuticals. We calculated age- and sex-standardized mean utilization rates in each domain and compared them with the corresponding estimates derived from official statistics and independent statutory health insurance data.

Results

The KORA-Age study underestimated the use of long-term care services (?52%), in-hospital days (?21%) and physician visits (?70%). In contrast, the assessment of drug consumption by postal self-report questionnaires yielded similar estimates to the analysis of insurance claims data (?9%).

Conclusion

Survey estimates based on self-report tend to underestimate true health care utilization in the older population. Direct validation studies are needed to disentangle the impact of recall and non-response bias.     

Occupational exposure to airborne mercury during gold mining operations near El Callao, Venezuela

Objective: The National Institute for Occupational Safety and Health (NIOSH) recently conducted a cross–sectional study during gold mining operations near El Callao, Venezuela. The purpose of the study was to assess mercury exposures and mercury-related microdamage to the kidneys. The study consisted of concurrent occupational hygiene and biological monitoring, and an examination of the processing techniques employed at the different mining facilities. Mercury was used in these facilities to remove gold by forming a mercury-gold amalgam. The gold was purified either by heating the amalgam in the open with a propane torch or by using a small retort. Methods: Thirty-eight workers participated in this study. Some participants were employed by a large mining company, while others were considered “informal miners” (self-employed). Mercury exposure was monitored by sampling air from the workers' breathing zones. These full-shift air samples were used to calculate time-weighted average (TWA) mercury exposure concentrations. A questionnaire was administered and a spot urine sample was collected. Each urine sample was analyzed for mercury, creatinine, and N-acetyl-ß-d-glucosaminidase (NAG). Results: The range for the 8-h TWA airborne mercury exposure concentrations was 0.1 to 6,315?μg/m³, with a mean of 183?μg/m³. Twenty percent of the TWA airborne mercury exposure measurements were above the NIOSH recommended exposure limit (REL) of 50?μg/m³, and 26% exceeded the American Conference of Governmental Industrial Hygienists (ACGIH) threshold limit value (TLV) of 25?μg/m³. The mean urine mercury concentration was 101?μg/g creatinine (μg/g-Cr), and the data ranged from 2.5 to 912?μg/g-Cr. Forty–two percent of the study participants had urine mercury concentrations that exceeded the ACGIH biological exposure index (BEI) of 35?μg/g-Cr. Urinary NAG excretion is considered a biological marker of preclinical, nonspecific microdamage to the kidney's proximal tubule cells. The mean urine NAG concentration was 3.6 International Units/g-Cr (IU/g-Cr) with a range of 0.5 to 11.5?IU/g-Cr. Three workers had urine NAG levels in excess of the reference values. Correlation analyses found statistically significant correlations between airborne mercury exposure and urine mercury level (P=0.01), and between urine mercury level and urine NAG excretion (P=0.01). In addition, the airborne mercury exposure data and urine mercury data were segregated by job tasks. A Wilcoxon rank sum test revealed significant correlations between tasks and mercury exposure (P=0.03), and between tasks and urine mercury level (P=0.02). Conclusions: The tasks with the highest mean airborne mercury exposures were “burning the mercury-gold amalgam” and “gold refining/smelting”. Recommendations were provided for improving the retort design to better contain mercury, for ventilation in the gold shops, and for medical surveillance and educational programs.     

Assessing surrogacy of data sources for institutional comparisons

Can administrative claims data, Z, serve as a surrogate for better clinical data, X, when assessing institutional performance? We consider an analysis of I hospitals, each of which involves an adjusted outcome. In the i ^th hospital, we denote the true association between the outcome and the risk factors using one data source by θ _i (X), the true association between the outcome and the risk factors using the other data source by γ _i (Z), and assume we have estimates of each available. Within hospital i, the estimated association parameters are jointly normally distributed such that conditional on γ _i (Z), a simple linear relationship exists between θ _i (X) and γ _i (Z). Methods are illustrated using mortality rates for 181,032 elderly US heart attack patients treated at 4322 hospitals. We find a strong linear relationship between the hospital standardized mortality rates adjusted by risk factors found in administrative claims data and rates adjusted by risk factors found in medical charts (posterior mean [95% interval] for slope: 0.997 [0.965,1.028]). However, the absolute and relative differences between the two sets of rates increase as hospital volume increases. For typically-sized standard deviations of claims-based rates, there is reasonable certainty of quality problems when the hospital’s claims-based rate is 0.72 times or smaller than the national mean or 1.45 times or greater than the national mean. Fewer hospitals are classified as either low-mortality or high-mortality hospitals when using claims-based estimates compared to chart-based estimates.     

Measuring fatigue in persons with multiple sclerosis: creating a crosswalk between the Modified Fatigue Impact Scale and the PROMIS Fatigue Short Form

Purpose

To create cross-walk tables to associate scores for the Modified Fatigue Impact Scale (MFIS) with scores for the Patient Reported Outcome Measurement Information System (PROMIS) Fatigue Short Form (SF) in persons with Multiple Sclerosis (MS).

Methods

Cross-walk tables were created using equipercentile linking and based on data collected at one time point in a longitudinal study of persons with MS (N?=?458). Validation of the tables was conducted using data collected at a subsequent time point (N?=?444). Deviations between estimates and actual scores were compared across levels of fatigue. The impact of sample size on the precision of sample mean estimates was evaluated using bootstrapping.

Results

Correlations between deviations and fatigue level for the PROMIS Fatigue SF and MFIS were (?0.31) and (?0.30), respectively, indicating moderately greater deviations with lower fatigue scores. Estimated sample means were impacted by sample size.

Conclusions

Cross-walk tables allow data from studies using different measures of fatigue to be combined to achieve larger sample sizes and to compare results. These tables are valid for group-level analyses with sample sizes of 150 or greater.     

Sexually dimorphic patterns of nutritional intake and eating behaviors in community-dwelling older adults with normal and slow gait speed

Objectives

Assess sex-specific nutritional intake and dietary habits of independently living older adults with normal and slow gait speeds.

Design

New Mexico Aging Process Study, cross-sectional, secondary data analysis.

Setting

Albuquerque, New Mexico USA.

Participants

Three-hundred fifteen adults 60 years and older (194 women and 121 men).

Measurements

Gait speed test, 3-day diet records, Mini-Mental State Examination, and body mass index.

Results

Slow gait speed was associated with lower total calories (?154 kcal/day) and zinc (1mg/day) (.05 < p <.1). Slower men consumed less protein (?4.1g/day), calcium (?140mg), fiber (-2.8g/day) and iron (?2.5mg/day) (p<.05). Slower women consumed less, protein (?5.5g/day), carbohydrate (?19.1g/day), fiber (?2.7gm/day), vitamin C (?18.4mg/day) and higher fat intake (p=0.03). Slower women snacked less, had trouble chewing/biting, and lived alone (p=.04). Slower men were less likely to snack.

Conclusions

We found sex-specific nutritional differences associated with gait speed. Those presenting with slow gait speed may need encouragement to increase meat and whole grain breads/cereal. Those with trouble eating should be advised on adapting diet to maintain adequate nutrition and encouraged on regular snacking to achieve higher nutrient intake. Prospective and randomized controlled studies are needed to confirm these findings and provide further evidence for putting these suggestions into practice.     

NON-ITERATIVE ROBUST ESTIMATORS OF VARIANCE COMPONENTS IN WITHIN-SUBJECT DESIGNS

Classic estimators of variance components break down in the presence of outliers and perform less efficiently under non-normality. In this article I present simple non-iterative estimators of variance components that are resistant to outliers and robust to systematic departures from normality, such as heavy tailedness of the distribution of responses. The proposed estimators are based on a robust extension of Hocking's AVE approach and are thus called RAVE estimators. I present results from a Monte Carlo comparison of RAVE versus classic estimation methods including maximum likelihood (ML), restricted maximum likelihood (REML) and minimum variance quadratic unbiased estimation (MIVQUE). Under simulated deviations from normality, RAVE estimators are associated with smaller mean squared errors than all the comparators, and in the normal case they exhibit a minimal loss in relative efficiency. A numerical example illustrates the proposed methodology. © 1997 John Wiley & Sons, Ltd.     

Temporal association patterns and dynamics of amyloid-β and tau in Alzheimer’s disease

The elusive relationship between underlying pathology and clinical disease hampers diagnosis of Alzheimer’s disease (AD) and preventative intervention development. We seek to understand the relationship between two classical AD biomarkers, amyloid-β_1?42 (Aβ_1?42) and total-tau (t-tau), and define their trajectories across disease development, as defined by disease onset at diagnosis of mild cognitive impairment (MCI). Using longitudinal data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), we performed a correlation analysis of biomarkers CSF Aβ_1?42 and t-tau, and longitudinal quantile analysis. Using a mixed effects model, with MCI onset as an anchor, we develop linear trajectories to describe the rate of change across disease development. These trajectories were extended through the incorporation of data from cognitively normal, healthy adults (aged 20–62 years) from the literature, to fit sigmoid curves by means of non-linear least squares estimators, to create curves encompassing the 50 years prior to MCI onset. A strong right-angled relationship between the biomarkers Aβ_1-42 and t-tau is detected, implying a highly non-linear relationship. The rate of change of Aβ_1-42 is correlated with the baseline concentration per quantile, reflecting a reduction in the rate of loss across disease within subjects. Regression models reveal significant amyloid loss relative to MCI onset (? 2.35 pg/mL/year), compared to minimal loss relative to AD onset (? 0.97 pg/mL/year). Tau accumulates consistently relative to MCI and AD onset, (2.05 pg/mL/year) and (2.46 pg/mL/year), respectively. The fitted amyloid curve shows peak loss of amyloid 8.06 years prior to MCI diagnosis, while t-tau exhibits peak accumulation 14.17 years following MCI diagnosis, with the upper limit not yet reached 30 years post diagnosis. Biomarker trajectories aid unbiased, objective assessment of disease progression. Quantitative trajectories are likely to be of use in clinical trial design, as they allow for a more detailed insight into the effectiveness of treatments designed to delay development of biological disease.     

Ilex paraguariensis (yerba mate) improves endocrine and metabolic disorders in obese rats primed by early weaning

Purpose

We showed that early weaned rats developed obesity, hyperleptinemia, leptin and insulin resistance at adulthood. Here, we studied the potential beneficial effects of Ilex paraguariensis aqueous solution upon body composition, glycemia, lipid and hormonal profiles, leptin signaling and NPY content.

Methods

To induce early weaning, lactating rats’ teats were blocked with a bandage to interrupt lactation during the last 3 days (EW group), while control offspring had free access to milk throughout lactation (C group). In postnatal day (PN) 150, EW offspring were subdivided into: EW and EW+ mate groups treated, respectively, with water or yerba mate aqueous solution (1 g/kg BW/day, gavage) during 30 days. C offspring received water for gavage. In PN180, offspring were killed.

Results

EW+ mate group presented lower body weight (?10 %), adipose mass (retroperitoneal:?40 % and epididymal:?44 %), total body fat (?43 %), subcutaneous fat (?46 %), visceral adipocyte area (?21 %), triglyceridemia (?31 %) and hypothalamic NPY content (?37 %) compared to EW group. However, hyperglycemia and lower HDL-c levels observed in EW group were not reverted with mate treatment. Although the hyperleptinemia, lower hypothalamic JAK2 and pSTAT3 content of EW group were not corrected by mate treatment, the hyperphagia and higher hypothalamic SOCS-3 content were normalized in EW+ mate group, indicating that the central leptin resistance could be restored.

Conclusion

Thus, the therapy with yerba mate solution was capable to reverse abdominal obesity, leptin resistance and hypertriglyceridemia, suggesting an important role of this bioactive component in the management of obesity in this programming model.     

Intake of vegetables and fruit and risk of esophageal adenocarcinoma: a meta-analysis of observational studies

Objectives

To study the association between the intake of fruit and vegetables and risk of esophageal adenocarcinoma (EAC), we summarized the evidence from observational studies in categorical and linear dose–response meta-analyses.

Methods

Eligible studies published up to June 2013 were retrieved via computerized searches of MEDLINE and EMBASE. Random-effects models were used to calculate summary relative risks (SRRs) and the corresponding 95 % confidence intervals (CIs). Between-study heterogeneity was assessed using the Cochran’s Q and I ² statistics.

Results

A total of 12 studies involving 1,572 cases of EAC were included in this meta-analysis. Based on the highest versus lowest analysis, inverse associations were observed between intakes of vegetable (SRRs = 0.76, 95 % CIs 0.59–0.96; P _{heterogeneity} = 0.098, I ² 40.4 %; n = 9 studies), intakes of fruit (SRRs = 0.73, 95 % CIs, 0.55–0.98; P _{heterogeneity} = 0.03, I ² = 52.9 %; n = 9 studies), and intakes of total vegetables and fruit combined (SRRs = 0.68, 95 % CI 0.49–0.93; P _{heterogeneity} = 0.162, I ² = 38.9 %; n = 5 studies). Similar results were also observed in a linear dose–response analysis.

Conclusion

These data support the hypothesis that intakes of vegetables and fruit may significantly reduce the risk of EAC. Further investigation with prospective designs, validated questionnaires, and good control of important confounders is warranted.     

Exposure data from multi-application, multi-industry maintenance of surfaces and joints sealed with asbestos-containing gaskets and packing

Fluid sealing devices (gaskets and packing) containing asbestos are manufactured and blended with binders such that the asbestos fibers are locked in a matrix that limits the potential for fiber release. Occasionally, fluid sealing devices fail and need to be replaced or are removed during preventive maintenance activities. This is the first study known to pool over a decade's worth of exposure assessments involving fluid sealing devices used in a variety of applications. Twenty-one assessments of work activities and air monitoring were performed under conditions with no mechanical ventilation and work scenarios described as "worst-case" conditions. Frequently, the work was conducted using aggressive techniques, along with dry removal practices. Personal and area samples were collected and analyzed in accordance with the National Institute for Occupational Safety and Health Methods 7400 and 7402. A total of 782 samples were analyzed by phase contrast microscopy, and 499 samples were analyzed by transmission electron microscopy. The statistical data analysis focused on the overall data sets which were personal full-shift time-weighted average (TWA) exposures, personal 30-min exposures, and area full-shift TWA values. Each data set contains three estimates of exposure: (1) total fibers; (2) asbestos fibers only but substituting a value of 0.0035 f/cc for censored data; and (3) asbestos fibers only but substituting the limit of quantification value for censored data. Censored data in the various data sets ranged from 7% to just over 95%. Because all the data sets were censored, the geometric mean and geometric standard deviation were estimated using the maximum likelihood estimation method. Nonparametric, Kaplan-Meier, and lognormal statistics were applied and found to be consistent and reinforcing. All three sets of statistics suggest that the mean and median exposures were less than 25% of 0.1 f/cc 8-hr TWA sample or 1.0 f/cc 30-min samples, and that there is at least 95% confidence that the true 95th percentile exposures are less than 0.1 f/cc as an 8-hr TWA.